D&C GLug - Home Page

[ Date Index ] [ Thread Index ] [ <= Previous by date / thread ] [ Next by date / thread => ]

[LUG]Re: Possible hardware failure - how to diagnose?

 

Given the extra info where it appears that the CPU / Ram subsystems are
still actually responsive, Helen's diagnosis and advice appears to ring very
sound.

Some of the PC power supplies are modular and may even have labels on
modular power ports (e.g. Motherboard, SATA, GPU, Molex). If you have one of
these it may be worth checking that your power distribution from the modular
power supply is as labelled or documented, and that power is correctly
distributed. You could try powering down disconnecting some equipment and
seeing if the problem goes away.

But then if the power supply is old, failing, or just underrated for the
equipment you're powering then you're probably looking at replacing that
part (can be a pain, as sometimes in tighter cases this means the
motherboard has to come out before you can physically remove the PSU).

Always still a good idea to clear out the dust puppies in your system though
!!!

-----Original Message-----
From: Helen McCall <helen.wildnfree@xxxxxxxxx> 
Sent: 20 September 2024 07:27
To: list@xxxxxxxxxxxxx
Subject: [LUG]Re: Possible hardware failure - how to diagnose?

Dear Sebastian,


On Thu, 19 Sep 2024 21:25:18 +0100
Sebastian via list <list@xxxxxxxxxxxxx> wrote:

> There's another thing that is interesting - I came to my computer just 
> before writing this email to discover it locked up again, and the 
> magic Alt-SysRq code for 'reboot' actually worked! So maybe it is not 
> completely irresponsive after all.

Many decades of experience building my own computers suggests to me that
this Alt-SysRq responsiveness points to your power supply to the hdd being
marginal. Things to check are the output level of your power supply, whether
you might have connected too many power hungry components such as big
graphics cards, or are overclocking anything and so increasing the current
requirement. Also of course check the power connection to the hdd is good.
If your power supply is being intermittently asked to supply its peak
current output, then the result will be that your operating system will for
a brief moment lose communication with your hdd, and loses registration with
the filing system and so become unresponsive to anything except functions
coded in the linux kernel. The functions of your window manager are not
coded in the kernel, and so cease to be responsive.

So make sure that the peak total current requirement of your system is never
anywhere near the peak power output of your power supply. Of course this
might also point to a faulty power supply if you have applied Ohm's Law to
your design and it still goes dead at times.

Best wishes from an ancient computing engineer from the Sixties!

Helen McCall
--
The Mailing List for the Devon & Cornwall LUG
FAQ: https://www.dcglug.org.uk/faq/

-- 
The Mailing List for the Devon & Cornwall LUG
FAQ: https://www.dcglug.org.uk/faq/